PATC Course: Big Data Analytics

Data: 08/Feb/2016 Time: 09:00 - 11/Feb/2016 Time: 18:00

Place: The course will take place in
Barcelona Supercomputing Centre,
within the UPC Campus Nord premises.
Room VX208, Vertex building

Target group: Level: (All courses are designed for specialists with at least 1st cycle degree or similar background experience) INTERMEDIATE: for trainees with some theoretical and practical knowledge;

Cost: There is no registration fee. The attendees would need to cover the expenses for travel, accommodation and meals.

Primary tabs

Day 1 08/02:  Introduction (Vassil Alexandrov)
Session 1: 9:30am – 1pm
  1. Data Science current trends session will focus on results of the latest key studies both and Europe and the USA an in the area of Data Science and outline the major trends, findings and recommendations.
Coffee break 11:00- 11:30
  1. Data Science definitions and mathematical foundations introduction. 
While tackling Big Data problems in many cases elementary or standard statistical approaches fail. New research methods are required to be developed to tackle such problems. Therefore this session will focus key research methods and approaches for Data Science, ranging from theory creating and theory testing approaches to conceptual-analytical approaches and experimental ones, that are able to lead to discovering global properties on data  These will be mainly deterministic and hybrid (stochastic/deterministic) methods and algorithms.
Session 2: 2pm – 6pm
  1. This session will focus on several key methods and algorithms (both serial and parallel) that enable to discover global properties on data while dealing with Big Data:
    • Network Science
    • Multi Constrained and Multi-Objective Optimization
    • Examples of using the above approaches
  2. Examples using the above approaches and some hands-on exercise
Coffee break 16:00 – 16:30
  1. Social Simulation Applications (Josep Casanovas)
Day 2 09/02:
Session 1: 9:30am – 1pm Data sharing (Anna Queralt)
  1. In this session we will provide an overview on current Open Data and data sharing approaches.
Usually, when talking about Big Data, the emphasis is put on how to efficiently store and analyse huge amounts of data. However, only when data from independent sources is combined it is possible to gain insights that would be impossible to obtain by analysing each dataset separately. Thus, it is essential that data, either public or private, is shared so that researchers, students, app developers or citizens in general can extract as much value as possible from it.
Coffee break 11:00- 11:30
  1. Hands-on exercise
Session 2: 2pm – 5pm Data analytics with Apache Spark - part 1 (Mario Macias)
In the recent years, Apache Spark has emerged as one of the most promising technologies for large-scale data processing in a fast and general way, with “programmer-friendly” interfaces and official bindings for many of the most used languages (Java, Scala, Python and R), extensive documentation and development tools. In addition, overcomes other MapReduce engines by 10x to 100x in terms of performance. This course introduces Apache Spark, as well as some of its core libraries for data manipulation, machine learning, graph analytics, etc.
  1. Introduction to the core concepts of Apache Spark: RDDs and Basic Data Access.
  2. Hands on: get the most frequent term from a text.
  3. Processing semi-structured data with Spark SQL.
  4. Hands on: statistical processing from Data Sheets.
Coffee break 16:00 - 16:30
  1. Multidisciplinary research and data analytics: Smart Cities (Maria Cristina Marinescu)
Day 3 10/02
Session 1: 9:30am – 1pm Data analytics with Apache Spark - part 2 (Mario Macias)
  1. Machine learning with Spark ML.
Coffee break 11:00- 11:30
  1. Hands on: clustering images according to their tags.
Session 2: 2pm – 6pm (Jordi Torres)
  1. Hello World in TensorFlow
If you want to learn how to start to program Deep Neural Networks, working with TensorFlow is an excellent way to start. TensorFlow is a machine learning library, which aims to bring large-scale, distributed machine learning and deep learning to everyone, open-sourced last November by Google. This tutorial will takes you through the TensorFlow programming model one step at a time.
  1. Hands-on exercises: beginning with basic machine learning models before moving on to a deep neural network, you will try out programming concepts as you learn them.
Coffee break 16:00 – 16:30
Day 4 11/02:
Session 1: 9:30am – 1pm (Alberto Abello)
  1. Big Data Management
  2. Big Data has many definitions and facets, we'll pay attention to the problems we have to face to store it and how we can process it. More specifically, we'll focus on the Apache Hadoop ecosystem and its two basic components, namely HBase and MapReduce engine.
Coffee break 11:00- 11:30
  1. Hands-on exercise
Session 2: 2pm – 6pm  Big (Javier Espinosa)
  1. Data Visualisation
Data visualizations are everywhere and are more important than ever. From creating a visual representation of data points as part of an executive presentation, to showcasing progress, or visualizing concepts for customer segments, data visualizations are a critical and valuable tool in many different situations. When it comes to big data, weak tools with basic features do not cut it so specific techniques should be applied. This course will address different techniques for visualizing big data collections including a vision of the visualization process as a complex and greedy task and then as out of the box solution that can help to analyse and interpret big data collection.
Coffee break 16:00 – 16:30
  1. Hands-on exercise